Unconventional reserves usually have unconventional decline behaviors, in that they do not follow the traditional exponential or hyperbolic decline trends. Often, they exhibit a "b" value greater than 1.0, which if kept constant, implies they will recover infinite reserves. Obviously, b-parameter must ultimately become less than 1.0

The variation of-b-parameter with time for producing wells was studied and resulted in the following conclusions:

  • It is extremely difficult to determine-b-parameter uniquely because of the associated noise in the data.

  • Within the scatter of production data, many values of-b-parameter can give an acceptable production history match.

  • This non-uniqueness of b, coupled with the data scatter, presents an opportunity for developing a decline methodology that honors engineering judgement, yet remains both familiar and simple to use.

Because the focus of this study was on unconventional wells, reservoir models of multiple-fractured horizontal wells in shales were studied in detail. Production forecasts were generated, and the-b-parameter value derived at every point in time, b(t). Unexpected patterns of b(t) were observed. However, once the prevailing flow regimes were clearly understood, these patterns could be explained. This study showed that:

  • Clearly, b(t) was not constant; sometimes it was sigmoidal, and often it was multisigmoidal

  • When noise was added to the synthetic production forecasts to represent real production data, a much simpler representation of-b-parameter was possible, and resulted in acceptable matches of the noisy production history.

  • The noise in the data combined with the non-uniqueness of b, allowed us to apply a non-complex solution to a complex problem. In other words, in this situation, data scatter along with the forgiving nature of b, was "our friend".

The commonly used hyperbolic equation assumes that b-parameter is constant. We have maintained the simplicity of this equation, and have developed a 3-Segment Hyperbolic Decline procedure. This proposed method:

  • has the advantage of familiarity - it uses the well-known traditional hyperbolic equation.

  • allows the b-parameter to vary in a simple manner for modeling a complex flow situation.

  • is able to represent production history, within the tolerance of the noisy data.

The time of the transition segment is a function of the reservoir description, and requires further study.

This content is only available via PDF.
You can access this article if you purchase or spend a download.